Measurement Invariance of the LGBT Microaggressions Scale Among Latino/o/x Persons and non-Latino/o/x People of Color

Abstract

Purpose:

Given limited measurement of intersectional microaggressions among those who identify as lesbian, gay, bisexual, transgender, queer, or additional sexual or gender minority (LGBTQ+) identities and as people of color (POC), this study tested whether differences exist in the factor structure of the LGBT People of Color Microaggression Scale (LGBT-PCMS) between Latina/o/x and non-Latina/o/x POC groups.

Methods:

Using Amazon’s Mechanical Turk, a sample of 491 LGBTQ+ Latina/o/x people and non-Latina/o/x POC (mean age = 29.66 ± 7.60) completed an online survey in December 2018, assessing psychosocial variables, intersectional minority stress measured via the LGBT-PCMS, and perceived stress. Hierarchical confirmatory factor analyses (CFA) assessed the fit of the LGBT-PCMS, with its three subscales specified as first-order latent factors and intersectional minority stress as a second-order latent factor. Invariance testing compared a second-order factor structure of the LGBT-PCMS between LGBTQ+ Latina/o/x people (n = 244) and LGBTQ+ non-Latina/o/x POC (n = 247) groups. Structural equation modeling was used to examine convergent validity with the perceived stress subscales.

Results:

Findings from the second-order CFA showed that the LGBT-PCMS demonstrated acceptable fit across the full sample, LGBTQ+ Latina/o/x people, and LGBTQ+ non-Latina/o/x POC. Full invariance was demonstrated between LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC groups. No link between the LGBT-PCMS and perceived stress subscales emerged.

Conclusion:

The LGBT-PCMS was indistinguishable between LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC, supporting that the LGBT-PCMS can be used as a single measure of intersectional minority stress or separate subscales.

Keywords

intersectional health disparities intersectional minority stress LGBTQ+ Latino/a/x persons,LGBTQ+ non-Latino/a/x people of color,measurement invariance

Introduction

With the extension of the minority stress model to lesbian, gay, bisexual, transgender, queer, or additional sexual or gender minority (LGBTQ+) individuals who also identify as people of color (POC), research on LGBTQ+ POC health disparities has grown substantially.^1–7 This work emphasizes intersectional minority stress (IMS),⁶ with scholars calling for its measurement in health research and greater attention to these experiences among LGBTQ+ POC.^8–10 The Williams Institute estimates 2.3 million LGBTQ+ Latina/o/x individuals in the U.S.,¹¹ a number projected to rise with increasing LGBTQ+ identification in this population.^12,13 As a distinct group, LGBTQ+ Latina/o/x people face heightened marginalization and rising hate crimes,^14,15 contributing to adverse health outcomes.^16,17

As a social determinant of health, racial/ethnic discrimination plays a critical role in the relationship between stress and health for LGBTQ+ Latina/o/x people. They experience unique forms of subtle racial/ethnic discrimination, including assumptions about intelligence, pathologizing of cultural values/communication, treatment as second-class citizens, stereotypes about shared experiences, and invalidations due to accents.¹⁸ This also includes more blatant discrimination, such as being treated as not “American enough,” being exoticized/sexually fetishized, and assumptions of criminality/criminal behavior.¹⁹

Second, racialized anti-Latina/o/x sentiments and anti-immigrant policies create unique sources of stress for Latina/o/x people by directly targeting who they are.²⁰ Although not new,²¹ the inhumane immigration enforcement under the 47th presidential administration further brings to light the unique sociopolitical stressors that shape the lived realities and health of Latina/o/x people. To this end, these unique experiences of racial/ethnic discrimination could impact them differently compared with non-Latina/o/x POC.²²

LGBTQ+ Latina/o/x people face an even more distinct blend of IMS compared with non-LGBTQ+ Latina/o/x people. Their sociocultural experiences are intertwined with racial/ethnic discrimination from the dominant groups and from their LGBTQ+ peers; their socialization as a Latina/o/x person, including gender norm expectations (e.g., expectations of masculine behavior and the subordination of women and femininity due to cultural values of machismo/caballerismo and marianismo); negative perceptions of LGBTQ+ identities within their Latina/o/x communities; and shame that can arise from anti-LGBTQ+ religious messaging from their Latina/o/x communities and familial upbringing.²³ Acculturation, trauma, and immigration can also significantly shape the ways in which LGBTQ+ Latina/o/x individuals experience IMS.^24,25

Although sociocultural factors likely shape IMS among LGBTQ+ Latina/o/x individuals compared with non-Latina/o/x POC, no studies examine how these factors affect measurement, limiting the development of tailored IMS scales.²⁶ Despite this, the LGBT People of Color Microaggressions Scale (LGBT-PCMS) is increasingly used in Latina/o/x samples.^27–33 The LGBT-PCMS captures nuanced psychosocial experiences among LGBTQ+ POC,^27,34 yet its application lacks consistency: some studies sum its three subscales,^28,29,32,35 whereas others model IMS as a latent construct.^24,30,36 Evidence on its factor structure is also mixed, with some supporting a three-factor model³⁷ and others not.³⁸ This inconsistency hinders comparability and limits clear conclusions about the measure’s structure in research with LGBTQ+ Latina/o/x populations.^28–33,35

Few studies examine measurement invariance (MI) of the LGBT-PCMS among LGBTQ+ Latina/o/x versus non-Latina/o/x groups. Although not designed for adolescents, Caba et al. found the racism subscale fully invariant by race/ethnicity among youth aged 13–17.³⁹ They reported partial invariance for the heterosexism in communities of color and LGBT relationship racism subscales. Huynh et al. observed partial invariance by race/ethnicity among LGBTQ+ POC adults.³⁸ In testing the LGBTQ+ PCMS-2, they found overall invariance across race/ethnicity, although some fit indices were suboptimal.³⁸

Despite representing 20% of LGBTQ+ individuals,¹¹ no studies test MI of the LGBT-PCMS comparing Latina/o/x individuals with non-Latina/o/x POC. Caba et al.’s sample included many Latina/o/x participants but focused on adolescents,³⁹ whereas Huynh et al.’s adult Latina/o/x sample was small, including in the LGBTQ+ PCMS-2.^38,40 We argue that Latina/o/x populations experience distinct IMS shaped by racialized socialization,²³ potentially affecting measurement.

Objectives/Purpose

There is uncertainty regarding how the LGBT-PCMS measures IMS for LGBTQ+ Latina/o/x and LGBTQ+ non-Latina/o/x POC groups. The first aim of our study was to examine whether there is MI for the LGBT-PCMS between LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC (see Fig. 1). Our second aim was to examine the convergent validity between IMS and measures of general perceived stress.

FIG. 1.

Hierarchical Confirmatory Factor Analysis Modeled for the lesbian, gay, bisexual, transgender, queer, or additional sexual or gender minority Latina/o/x and non-Latina/o/x People of Color Samples and Measurement Invariance Testing. I, item number which corresponds with item number in Table 2. Dashed lines indicate marker indicators.

Methods

Participants

Participants were recruited in December 2018 using Amazon Mechanical Turk (MTurk). After consenting to participate, we asked participants to answer two screening questions to ensure they identified as LGBTQ+ Latina/o/x or LGBTQ + non-Latina/o/x POC; if they answered no to either question, they were directed to the end of the survey. Their race, ethnicity, sexual orientation, and gender identities were further confirmed when they answered questions after the demographics. Our final sample was comprised of 491 LGBTQ+ Latina/o/x persons and LGBTQ+ non-Latina/o/x POC aged 18–69 years (mean = 29.66, standard deviation = 7.60) after removing nine participants.

Procedure

Informed consent was obtained before the survey. After completion, participants were directed to MTurk for compensation of $0.10, consistent with other studies that require a similar amount of time.⁴¹ To filter bots, only participants with a task completion rate of 95% or higher were allowed to take the survey, as this yields better data quality.⁴² This study was reviewed and approved on November 29th, 2018 by the Institutional Review Board of the University of La Verne (Protocol #: 2018-94-CAS), where the second author held a previous academic appointment.

Measures

Demographic information. Participants filled out demographic information. This included relationship status, current human immunodeficiency virus (HIV) status, residential environment, ethnicity, race, sexual orientation, and gender identity (see Table 1).

Table 1.

Demographic Characteristics of the Full Sample (N = 491)

Variable	n (%)	Latina/o/xpersons n (%)	Non-Latina/o/x peopleof color n (%)
Ethnicity
Mexican/Mexican descent		97 (39.75)
Puerto Rican/Puerto Rican descent		35 (14.34)
South American/South American descent		31 (12.71)
Central American/Central American descent		30 (12.30)
Cuban/Cuban descent		26 (10.66)
Dominican/Dominican descent		23 (9.43)
More than one Latina/o/x ethnicity		2 (0.82)
Race
White European	102 (20.77)	102 (41.80)
Black/African American	236 (48.07)	75 (30.74)	161 (65.18)
Asian/Asian American	44 (8.96)	9 (3.69)	35 (14.20)
Native American or Alaska Native	36 (7.33)	25 (10.25)	11 (4.45)
Hawaiian/Pacific Islander	12 (2.44)	8 (3.28)	4 (1.62)
Multiracial	61 (12.42)	25 (10.25)	36 (14.57)
Sexual orientation
Straight/heterosexual^a	7 (1.43)	3 (1.23)	4 (1.62)
Gay	145 (29.53)	77 (31.56)	68 (27.53)
Bisexual	198 (40.33)	97 (39.75)	101 (40.89)
Lesbian	107 (21.79)	51 (20.90)	56 (22.67)
Mixed/fluid sexual orientation	20 (4.07)	11 (4.51)	9 (3.64)
Other sexual orientations^b	14 (2.85)	5 (2.05)	9 (3.64)
Gender identity
Cisgender man/male	206 (41.96)	107 (43.85)	99 (40.08)
Cisgender woman/female	216 (43.99)	98 (40.16)	118 (47.77)
Trans man/male	24 (4.89)	14 (5.74)	10 (4.05)
Trans woman/female	15 (3.05)	8 (3.28)	7 (2.83)
Genderqueer/gender nonconforming	28 (5.70)	17 (6.97)	11 (4.45)
Other gender identity^c	2 (0.41)	0 (0)	2 (0.81)
Relationship status
Single	195 (39.71)	91 (37.30)	104 (42.11)
Monogamous	158 (32.18)	76 (31.15)	82 (33.20)
Open	107 (21.79)	58 (23.77)	49 (19.84)
Polyamorous	26 (5.30)	17 (6.97)	9 (3.64)
Other	5 (1.02)	2 (0.82)	3 (1.21)
HIV status
Negative	399 (81.26)	181 (74.18)	218 (88.26)
Positive	61 (12.42)	39 (15.98)	22 (8.91)
Don’t know	31 (6.31)	24 (9.84)	7 (2.83)
Residential environment
Rural	68 (13.85)	40 (16.39)	28 (11.34)
Suburban	214 (43.58)	98 (40.16)	116 (46.96)
Urban	209 (42.57)	106 (43.44)	103 (41.70)

This included transgender men, transgender women, and gender nonconforming/genderqueer/nonbinary.

Other sexual orientations included asexual, demi-romantic, and pansexual.

Participants who indicated other gender identities identified as nonbinary. Percentages in this table have been rounded up to two exact decimal places and may not total exactly 100%.

Sexual orientation and gender identity. Sexual orientation and gender identity (SOGI) were assessed using the recommendations of The Williams Institute.⁴³ Sexual orientation was assessed via “Do you think of yourself as…” which allowed respondents to choose the following options: straight/heterosexual, gay, lesbian, bisexual, or other. We added one last sexual orientation option to our survey: mixed/fluid. Gender identity was assessed with “How do you describe your gender identity?” Possible options included cisgender man/male, cisgender woman/female, trans man/male, trans woman/female, genderqueer/gender nonconforming, or a different identity (e.g., nonbinary). If participants listed a different identity, they were asked to write this option as a text response (see Table 1).

Race and ethnicity. This variable was assessed in accordance with modified federal guidelines.⁴⁴ Participants were asked to self-report their ethnicity with the question “What is your ethnicity?” which had the following options: Not Latino/a/x or Latino/a/x. Next, participants were asked about their race via the item “What is your race? (Check all that apply)” and were given the following options: White European, Black/African American, Native American or Alaska Native, Hawaiian or Pacific Islander, and Asian/Asian American. Respondents who marked more than two racial groups were coded as Multiracial (see Table 1). For our analyses, individuals who identified as Latina/o/x, regardless of their racial background, were coded as Latina/o/x. Those who answered that they were not Latina/o/x were then coded as non-Latina/o/x POC.

LGBT-PCMS. The LGBT-PCMS is an 18-item self-report measure of IMS assessed by three subscales: (1) LGBT racism, (2) Heterosexism in communities of color, and (3) LGBT relationship racism.²⁷ Items are rated on a 6-point Likert scale (0 [Did not happen/Not applicable to me] to 5 [It happened, and it bothered me Extremely]). The total score and three subscales had adequate reliability (see Table 2).

Perceived stress. This was measured using the 10-item Perceived Stress Scale (PSS-10).⁴⁵ We used the two subscales (perceived helplessness and self-efficacy) from the PSS-10, which have shown validity and reliability in Latina/o/x persons,⁴⁶ compared to the full PSS-10.⁴⁷ Items are scored on a five-point Likert scale (0 [Never] to 4 [Very Often]). Reliability was adequate (perceived helplessness α = 0.87 and self-efficacy α = 0.78).

Table 2.

Reliability Estimates and Confirmatory Factor Analysis of the Full Sample, Stratified by Lesbian, Gay, Bisexual, Transgender, Queer, or Additional Sexual or Gender Minority Latina/o/x and Non-Latina/o/x People of Color Groups

	Full sample (n = 491)	LGBTQ+ Latina/o/x group (n = 244)	LGBTQ+ non-Latina/o/x people of color group (n = 247)
Variable	λ (SE)	λ (SE)	λ (SE)
Factor 1: LGBT racism	α = 0.896ω = 0.898	α = 0.920ω = 0.902	α = 0.930ω = 0.914
11. Not being able to trust White LGBT people	0.733^a	0.696^a	0.765^a
7. Feeling misunderstood by White LGBT people	0.826 (0.048)	0.801 (0.077)	0.852 (0.061)
16. Having to educate White LGBT people about race issues	0.815 (0.063)	0.818 (0.107)	0.808 (0.076)
13. Being the token LGBT person of color in groups of or organizations	0.733 (0.068)	0.793 (0.111)	0.677 (0.085)
9. Being told that “race isn’t important” by White LGBT people	0.767 (0.076)	0.737 (0.126)	0.804 (0.094)
17. White LGBT people saying things that are racist	0.753 (0.073)	0.733 (0.112)	0.776 (0.100)
Factor 2: Heterosexism in communities of color	α = 0.877ω = 0.877	α = 0.907ω = 0.888	α = 0.899ω = 0.883
5. Not being accepted by other people of your race/ethnicity because you are LGBT	0.770^a	0.785^a	0.756^a
18. Feeling misunderstood by people in your ethnic/racial community	0.757 (0.035)	0.785 (0.051)	0.730 (0.047)
10. Feeling invisible because you are LGBT	0.702 (0.047)	0.722 (0.063)	0.686 (0.070)
1. Difficulty finding friends who are LGBT and from your racial/ethnic background	0.708 (0.053)	0.685 (0.069)	0.733 (0.081)
4. Feeling unwelcome at groups or events in your racial/ethnic community	0.771 (0.044)	0.771 (0.056)	0.772 (0.069)
14. Not having any LGBT people of color as positive role models	0.709 (0.052)	0.727 (0.064)	0.695 (0.085)
Factor 3: LGBT relationship racism	α = 0.871ω = 0.873	α = 0.902ω = 0.875	α = 0.915ω = 0.891
3. Being rejected by other LGBT people of your same race/ethnicity	0.663^a	0.655^a	0.671^a
6. Being rejected by potential dating or sexual partners because of your race/ethnicity	0.721 (0.078)	0.722 (0.104)	0.723 (0.114)
12. Being seen as a sex object by other LGBT people because of your race/ethnicity	0.749 (0.083)	0.746 (0.114)	0.751 (0.118)
15. Reading dating app profiles that say “White people only”^b	0.695 (0.093)	0.661 (0.129)	0.727 (0.132)
2. Feeling like White LGBT people are only interested in you for your appearance	0.769 (0.090)	0.755 (0.125)	0.781 (0.130)
8. Being discriminated against by other LGBT people of color because of your race	0.788 (0.079)	0.784 (0.106)	0.790 (0.115)
Second-order factor: Intersectional minority stress	ω_L2 = 0.919ω_L1p = 0.947	ω_L2 = 0.922ω_L1p = 0.947	ω_L2 = 0.915ω_L1p = 0.948
LGBT racism	0.875^a	0.901^a	0.847^a
Heterosexism in communities of color	0.885 (0.083)	0.886 (0.125)	0.881 (0.115)
LGBT relationship racism	0.913 (0.092)	0.899 (0.120)	0.930 (0.144)

λ, factor loadings; SE, robust standard error, α, internal consistency for the latent factor; ω, additional reliability metric for latent factor; ω_L2, reliability metric for second-order latent factor; ω_L1p, reliability metric that measures proportion of observed variance explained by the second-order factor after partialling the uniqueness from the first-order factor; all factor loadings are statistically significant.

^aMarker indicator.

^bItem adjusted for relevancy.

LGBTQ+, lesbian, gay, bisexual, transgender, queer, or additional sexual or gender minority.

Data analysis

All analyses were conducted using R.⁴⁸ Listwise deletion was used to account for missing data. Hierarchical confirmatory factor analyses (CFAs) were conducted to examine the LGBT-PCMS factor structure for both the full sample and the LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC samples. Figure 1 presents the CFA models tested using maximum likelihood with robust standard errors (MLR) estimation, where the LGBT-PCMS is referred to as IMS as the second-order latent factor and the three subscales are represented as first-order latent factors.

Although Balsam et al. treated items as ordinal,²⁷ MLR was used as a robust version of the most commonly used estimation method and the default for several software packages while yielding similar findings to ordinal estimation methods,⁴⁹ particularly with five or more response options.^50,51 We used recommendations from Putnick and Bornstein, including comparative fit index (CFI; acceptable > 0.90; good > 0.95), root mean square error of approximation (RMSEA; unacceptable > 0.10; good fit < 0.06), and standardized root mean squared residual (SRMR; good < 0.05) for the CFA and structural equation modeling (SEM).⁵²

Comparisons of the LGBT-PCMS hierarchical factor structure between LGBTQ+ Latina/o/x persons and LGBTQ+ non-Latina/o/x POC were tested through MI. Constraints at each level of MI (i.e., configural, metric, scalar, and strict invariance) were tested sequentially, where partial invariance was established if invariance was not met.^52,53 Latent means were compared between the LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC groups, with effect sizes being quantified using Cohen’s d. Each step of invariance testing was evaluated by assessing the chi-square change (Δχ²) for statistical significance relative to the change in degrees of freedom in model comparisons,⁵³ along with changes in CFI (metric = ≤ 0.010, scalar/strict = ≤ 0.010), RMSEA (metric = ≤ 0.015, scalar/strict = ≤ 0.015), and SRMR (metric = ≤ 0.030, scalar/strict = ≤ 0.010) indicating invariance.⁵⁴

To test convergent validity, structural equation modeling (SEM) was used to examine the associations of IMS and the PSS-10 subscales. These associations were compared between LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC using multigroup SEM. Group differences were detected by a significant chi-square test between the freely estimated model and the model constraining the intercept and regression paths.

Although not the aim of this study, we conducted exploratory analyses using a bifactor CFA to examine the factor structure and MI for SOGI groups. For the sexual orientation groups, we compared groups based on sexual orientation and sex assigned at birth. For invariance testing, we compared gay men with bisexual men and lesbian women with bisexual women. For the gender identities, we compared cisgender men, cisgender women, and transgender individuals (see Supplementary Data and Supplementary Tables S1, S2, S3, S4, S5, S6, and S7).

Results

To test for a hierarchical structure of the LGBT-PCMS in the full sample, an initial CFA was performed, which showed acceptable fit: χ2(132) = 351, p < 0.001; CFI = 0.943; RMSEA = 0.067, 90% confidence interval (CI): 0.059–0.076; SRMR = 0.043. Model fit was good for the LGBTQ + Latina/o/x group (χ²(132) = 210, p < 0.001, CFI = 0.959, RMSEA = 0.056, 90% CI: 0.041–0.07, SRMR = 0.041) and acceptable for the LGBTQ + non-Latina/o/x POC group (χ²(132) = 268, p < 0.001, CFI = 0.931, RMSEA = 0.075, 90% CI: 0.062–0.088, SRMR = 0.057). The factor loadings and reliability estimates revealed a hierarchical structure with acceptable fit for the first- and second-order latent factors (see Table 2).

Testing the bifactor CFA indicated a better fit than the hierarchical CFA (χ²(15) = 40.28, p < 0.001) for the full sample. However, there were large standard errors, “vanishing” of the LGBT relationship racism subscale variance (ᴪ = 0.003, p = 0.925), and non-convergence (see Supplementary Data). This suggests model under-identification.

Table 3 presents the fit for each level of MI testing, along with the differences in model fit indices. There were no statistically significant differences between MI levels when comparing LGBTQ+ Latina/o/x persons and LGBTQ+ non-Latina/o/x POC groups. When comparing latent means, no statistically significant differences were found between the two groups, with negligible effect sizes observed (see Table 4).

Table 3.

Measurement Invariance Findings Between Lesbian, Gay, Bisexual, Transgender, Queer, or Additional Sexual or Gender Minority Latina/o/x and Non-Latina/o/x People of Color Groups

Model	χ²(df)	CFI	RMSEA (90% CI)	SRMR	Δχ^2a(df)	ΔCFI	ΔRMSEA	ΔSRMR
Configural	479.16 (264)	0.95	0.066 (0.057–0.075)	0.046	—	—	—	—
Metric	503.66 (281)	0.94	0.064 (0.055–0.073)	0.052	20.01 (17)	−0.000	−0.002	0.006
Scalar	518.97 (295)	0.94	0.063 (0.054–0.072)	0.053	13.45 (14)	0.000	−0.002	0.001
Strict	534.94 (313)	0.95	0.061 (0.052–0.069)	0.053	15.80 (18)	0.001	−0.002	0.000

Difference in chi-square uses standard test statistics, not the robust chi-square used for model fit.

CFI, comparative fit index; df, degrees of freedom; RMSEA, root mean square error of approximation; SRMR, standardized root mean squared residual.

Table 4.

Means, Standard Deviations, and Effect Size in Latent Means Between Lesbian, Gay, Bisexual, Transgender, Queer, or Additional Sexual or Gender Minority Latina/o/x and Non-Latina/o/x People of Color Groups

Variable	LGBTQ+ Latina/o/x group (n = 244)	LGBTQ+ non-Latina/o/x POC group (n = 247)	Cohen’s d
Scalar invariance	M (SD)	M (SD)
LGBT racism	0.00 (0.72)	0 (0)	0
Heterosexism in communities of color	−0.03 (0.76)	0 (0)	0.05
LGBT relationship racism	0.01 (0.68)	0 (0)	0.03^a
Intersectional minority stress	−0.01 (0.96)	0 (0)	0.02
Strict invariance
LGBT racism	0.00 (0.72)	0 (0)	0.01
Heterosexism in communities of color	−0.03 (0.75)	0 (0)	0.06
LGBT relationship racism	0.01 (0.68)	0 (0)	0.02^a
Intersectional minority stress	−0.01 (0.96)	0 (0)	0.01

Means were set to zero for the non-Latina/o/x POC group for latent mean comparisons.

Absolute value of d is reported instead of a negative value for ease of interpretation.

M, mean; POC, people of color; SD, standard deviation.

Convergent validity was not attained, as IMS was not associated with perceived helplessness (b = −0.03, p = 0.912) or self-efficacy (b = −0.01, p = 0.967), although model fit was acceptable in the full sample (χ²(166) = 391.41; CFI = 0.944; RMSEA = 0.060, 90% CI 0.052–0.067; SRMR = 0.040). Differences were not found between the LGBTQ + Latina/o/x and the LGBTQ + non-Latina/o/x groups (χ²(18) = 13.91, p = 0.735).

Discussion

Our study investigated whether scores on the LGBT-PCMS work similarly for LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC. These findings support the factorial validity of the LGBT-PCMS’ three-factor structure,²⁷ while extending IMS as a second-order factor in both groups. MI findings showed no differences between LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC. We found that IMS was not linked to perceived stress.

From the CFA and corresponding MI analyses, the LGBT-PCMS showed acceptable fit in the full sample and in LGBTQ+ Latina/o/x and LGBTQ+ non-Latina/o/x groups, consistent with previous literature.^27,37 The second-order factor was tested due to evidence of IMS being measured through a single construct rather than three subscales. Although limited, the second-order and general factors from the hierarchical CFA and bifactor CFA showed evidence for future practitioners and researchers to use IMS as the construct for this scale. This is reflected by previous literature indicating strong correlations between the LGBT-PCMS subscales (0.71–0.83)³⁷ and “vanishing” of variation when testing the bifactor CFA.

Furthermore, the LGBT-PCMS showed acceptable fit for LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC with no differences across all levels of MI testing. What makes our findings unique is that we adjusted for previously overlooked factors in the published literature, including inappropriate estimation methods,³⁷ matching indicators to incorrect latent variables,³⁸ and testing latent variables separately.³⁹

We also examined latent means for the IMS and the corresponding subscales. The findings showed that they did not differ, suggesting that LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC groups similarly experience these constructs. Previous MI studies with this measure in adult samples found partial invariance of the LGBT-PCMS within racial/ethnic groups,³⁸ some demonstrated varying levels of fit in their MI testing in adults,⁴⁰ whereas corrections were needed for the LGBT-PCMS in adolescents to attain partial invariance.³⁹ In contrast, our findings showed full invariance comparing LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC.

Lack of differences between LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC at the latent mean level suggests that these groups experience IMS similarly. These findings suggest that future analyses could analyze IMS collectively for these groups or compare these groups through composite scores to test associations within the minority stress model, such as with physical and mental health outcomes. Perhaps the LGBT-PCMS does not comprehensively capture the differences that may exist between these groups, such as heterosexism and cissexism in familial contexts, as measured by the new LGBTQ+ PCMS-2.⁴⁰ However, further validation of this measure is sorely needed.

Regarding convergent validity, our findings revealed no association between IMS and measures of perceived stress. This is inconsistent with previous literature on the associations of the LGBT-PCMS with psychological adjustment measures.^27,37 The difference in findings may be the result of replicating the association of IMS and perceived stress²⁷ through the PSS-10 rather than the PSS-4.⁴⁷ The use of the PSS-10, including its subscales, to replicate the work of Balsam et al. who utilized the PSS-4, indicates that the PSS-10 serves as a comprehensive measure of perceived stress; this could explain the lack of evidence for convergent validity.^47,55 Moreover, it could also be argued that perceived stress is an entirely different construct than IMS, in which case our null findings regarding convergent validity support this conjecture. Nonetheless, future studies should utilize appropriate mental health constructs to better test the minority stress model,⁵ such as depressive symptoms, to parse out the associations between general stressors (i.e., perceived stress) and proximal and distal minority stressors (i.e., IMS).⁵⁶

Practical implications

Findings from this study have implications for researchers and practitioners working with LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC. The LGBT-PCMS can be used with these groups to assess IMS. However, given null associations, clinicians must assess IMS and stress for LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC separately, as these distinct constructs could have independent effects on LGBTQ+ Latina/o/x and LGBTQ+ non-Latina/o/x POC health, in line with evidence-based clinical assessment.⁵⁷

Limitations

First, the sample sizes for our LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC groups were not as large as those in one invariance study³⁹; however, our groups were larger than those of Huynh et al.⁴⁰ Nevertheless, sample size is often not a significant factor in MI studies.undefined⁵³ Second, although precautions were taken,⁴² recruiting from MTurk may result in non-representative samples that may conceivably not reflect our population of interest. Our sample was mostly cisgender sexual minority men and women of color; yet, qualitative differences in IMS exist for transgender POC,⁵⁸ with some indicating that the LGBT-PCMS does not reflect transgender POC IMS.⁵⁹ More research is needed to explore SOGI differences in the LGBT-PCMS (see Supplementary Data and Supplementary Tables S1, S2, S3, S4, and S5).

We used perceived stress as a measure of general psychological distress rather than a construct more closely aligned with IMS to examine convergent validity, in line with others.³² These data were collected in 2018, reflecting a different time compared with the current sociopolitical climate that shows increased hostility and IMS for LGBTQ+ Latina/o/x people and LGBTQ+ non-Latina/o/x POC.

Our assessment of ethnicity does not fully capture the complexities of Latina/o/x ethnic identities⁶⁰; however, as a proxy measure for racial socialization, it allowed us to focus solely on whether differences that could be attributed to Latina/o/x culture could influence how IMS is experienced. Comparing Latina/o/x as an additional race/ethnicity category may also reduce reporting of White and Other races with the trade-off of reducing Black and Native American or Alaska Native reporting among Latina/os/x people⁶¹; however, this study’s null findings provide further evidence to support examining Latina/o/x people as their own distinct group, similar to other MI studies in this area of research.^38,40

Lastly, although our bifactor CFA was a significantly better-fitting model, it was under-identified. Future research is needed to examine item-level responses in other models to assess if they function differently across LGBTQ+ POC groups.

Conclusions

Our findings indicate that the LGBT-PCMS is a valid, reliable measure for LGBTQ+ Latina/o/x people and non-Latina/o/x POC. These results support its use in health research⁵⁵^,undefined; however, more attention is needed to evaluate the LGBT-PCMS and its updated versions in representative samples to further test validity and compare ethnic group differences. Accurate IMS measurement enables researchers/clinicians to assess the intersectional psychosocial stress pathways through which health disparities develop and are maintained among LGBTQ+ Latina/o/x people and non-Latina/o/x POC.

Ethical Considerations

This study was reviewed and approved on November 29^th, 2018, by the Institutional Review Board of the University of La Verne, Protocol #: 2018-94-CAS.

Consent to Participate

All participants provided their written consent to participate in this study.

Consent for Publication

All participants provided their written consent for the results of this study to be published.

Data Availability

De-identified data may be provided by reasonable request to the corresponding author.

Declaration of Conflicting Interest

The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ Contributions

J.A.P.: Conceptualization (equal), formal analysis (lead), validation (lead), visualization (lead), writing original draft (equal), and writing—reviewing and editing (equal); J.J.G.-G.: Conceptualization (lead), data curation (supporting), formal analysis (supporting), investigation (lead), methodology (lead), project administration (lead), supervision (lead), validation (lead), writing original draft (equal), and writing—reviewing and editing (equal); M.J.W.: Conceptualization (supporting) and writing—review and editing (supporting); M.S.A.: Investigation (supporting) and writing—review and editing (supporting); N.N.N.: Writing—review and editing (supporting); D.G.S.: Writing—review and editing (supporting); A.H. Writing—review and editing (supporting); D.C.: Writing—review and editing (supporting).

Footnotes

Acknowledgments

The authors would like to thank the participants for taking part in this study.

Funding Statement

The authors received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

References

1. Barnett

, del Río-González

, Parchem

, et al. Content analysis of psychological research with lesbian, gay, bisexual, and transgender people of color in the United States: 1969–2018. Am Psychol 2019; 74: 898–911; doi: 10.1037/amp0000562

2. Ghabrial

and Ross

. Representation and erasure of bisexual people of color: A content analysis of quantitative bisexual mental health research. Psychol Sex Orientat Gend Divers 2018; 5: 132; doi: 10.1037/sgd0000286

3. Wade

and Harper

. Young Black gay/bisexual and other men who have sex with men: A review and content analysis of health-focused research between 1988 and 2013. Am J Mens Health 2017; 11: 1388–1405; doi: 10.1177/1557988315606962

4. Rich

, Salway

, Scheim

, et al. Sexual minority stress theory: Remembering and honoring the work of Virginia Brooks. LGBT Health 2020; 7: 124–127; doi: 10.1089/lgbt.2019.0223

5. Meyer

. Prejudice, social stress, and mental health in lesbian, gay, and bisexual populations: Conceptual issues and research evidence. Psychol Bull 2003; 129: 674–697; doi: 10.1037/0033-2909.129.5.674

6. Cyrus

. Multiple minorities as multiply marginalized: Applying the minority stress theory to LGBTQ people of color. J Gay Lesbian Ment Health 2017; 21: 194–202; doi: 10.1080/19359705.2017.1320739

7. García

. (ed.) Heart Brain and Mental Health Disparities for LGBTQ People of Color. Cham, Switzerland: Springer International Publishing, 2021.

8. Alvidrez

, Greenwood

, Johnson

, et al. Intersectionality in public health research: A view from the National Institutes of Health. Am J Public Health 2021; 111: 95–97; doi: 10.2105/AJPH.2020.305986

9. Sievwright

, Stangl

, Nyblade

, et al. An expanded definition of intersectional stigma for public health research and praxis. Am J Public Health 2022; 112: S356–S361; doi: 10.2105/AJPH.2022.306718

10.

10. Turan

, Elafros

, Logie

, et al. Challenges and opportunities in examining and addressing intersectional stigma and health. Focus (Am Psychiatr Publ) 2025; 23: 70–84; doi: 10.1176/appi.focus.25023006

11.

11.The Williams Institute. Race and well-being among LGBT adults. Los Angeles, CA, https://williamsinstitute.law.ucla.edu/visualization/lgbt-race-comparison/ [2022, accessed July 30, 2025].

12.

12. Jones

. Gallup. Growing LGBT ID seen across major U.S. racial, ethnic groups. Washington, DC, https://news.gallup.com/poll/393464/growing-lgbt-seen-across-major-racial-ethnic-groups.aspx [2022, accessed July 30, 2025].

13.

13.Public Religion Research Institute (PRRI). A political and cultural glimpse into America’s future: Generation Z’s views on generational change and the challenges and opportunities ahead. Washington, DC, https://www.prri.org/research/generation-zs-views-on-generational-change-and-the-challenges-and-opportunities-ahead-a-political-and-cultural-glimpse-into-americas-future/ [2024, accessed July 30, 2025].

14.

14. Papakyriakopoulos

, Zuckerman

. The media during the rise of Trump: Identity politics, immigration, “Mexican” demonization and hate-crime. Proc Int AAAI Conf Web Soc Media 2021; 15: 467–478; doi: 10.1609/icwsm.v15i1.18076

15.

15. Kutateladze

. Hate crime victimization and reporting within Miami’s queer Latine immigrant population. Law Hum Behav 2022; 46: 429–439; doi: 10.1037/hb0000505

16.

16. Lockwood

, Farrell

, Cuevas

, et al. Bias crime and victimization among Latinx adults: Formal and informal help seeking. J Interpers Violence 2022; 37: NP22600–NP22626; doi: 10.1177/08862605211072175

17.

17. Rhodes

, Mann-Jackson

, Alonzo

, et al. The health and well-being of Latinx sexual and gender minorities in the USA: A call to action. In: New and emerging issues in Latinx health. Cham, Switzerland: Springer International Publishing, 2019, pp. 217–236.

18.

18. Rivera

, Forquer

, and Rangel

. Microaggressions and the life experience of Latina/o Americans. In: Sue

. (ed.) Microaggressions and Marginality: Manifestation, Dynamics, and Impact. Hoboken, NJ: John Wiley & Sons, 2010, pp. 59–83.

19.

19. Nadal

, Mazzula

, Rivera

, and Fujii-Doe

. Microaggressions and Latina/o Americans: An analysis of nativity, gender, and ethnicity. J Latina/o Psychol 2014; 2: 67–78.

20.

20. Rojas Perez

, Silva

, Galvan

, et al. Buscando la calma dentro de la tormenta: A brief review of the recent literature on the impact of anti-immigrant rhetoric and policies on stress among Latinx immigrants. Chronic Stress (Thousand Oaks) 2023;7:24705470231182475; doi: 10.1177/24705470231182475

21.

21. Lopez

, Gonzalez-Barrera

and Krogstad

. More Latinos Have Serious Concerns About Their Place in America Under Trump. Washington, DC, https://www.pewresearch.org/hispanic/wp-content/uploads/sites/5/2018/10/PewResearch-Center_Latinos-have-Serious-Concerns-About-Their-Place-in-America_2018-10-25.pdf [2018, accessed October 19, 2025].

22.

22. Busse

, Yim

, Campos

. Social context matters: Ethnicity, discrimination and stress reactivity. Psychoneuroendocrinology 2017;83:187–193; doi: 10.1016/j.psyneuen.2017.05.025

23.

23. Schmitz

, Robinson

, Tabler

, et al. LGBTQ+ Latino/a young people’s interpretations of stigma and mental health: An intersectional minority stress perspective. Soc Ment Health 2020; 10: 163–179; doi: 10.1177/2156869319847248

24.

24. Zelaya

, Rosales

, García

, et al. Intersectional forms of racism and sexual minority stress are associated with alcohol use among Latinx sexual minority men with greater Hispanic/Latinx cultural orientation. Am J Drug Alcohol Abuse 2023; 49: 228–238; doi: 10.1080/00952990.2023.2176235

25.

25. Estrada

, Jones

, Rivera

, et al. Trauma and Latinx sexual- and gender-minority immigrants in the U.S. Curr Opin Psychol 2022;47:101439; doi: 10.1016/j.copsyc.2022.101439

26.

26. Torres

, Mata-Greve

, Bird

, et al. Intersectionality research within Latinx mental health: Conceptual and methodological considerations. J Latina/o Psychol 2018; 6: 304–317; doi: 10.1037/lat0000122

27.

27. Balsam

, Molina

, Beadnell

, et al. Measuring multiple minority stress: The LGBT People of Color Microaggressions Scale. Cultur Divers Ethnic Minor Psychol 2011; 17: 163–174; doi: 10.1037/a0023244

28.

28. Salerno

, Pease

, Gattamorta

, et al. Impact of racist microaggressions and LGBTQ-related minority stressors: Effects on psychological distress among LGBTQ+ young people of color. Prev Chronic Dis 2023;20:E63; doi: 10.5888/pcd20.220371

29.

29. Randelman

and Watson

. The relations among microaggressions, collective action, and psychological outcomes among sexual minority Latinx people. J Latinx Psychol 2023; 11: 240–257; doi: 10.1037/lat0000225

30.

30. Papa

and Parmenter

. At the intersection of intersectional identity and microaggressions: An examination of the experiences and identity of sexual and gender diverse BIPOC individuals. Cultur Divers Ethnic Minor Psychol 2025; 31: 175–186; doi: 10.1037/cdp0000624

31.

31. Parmenter

and Barrita

. A preliminary model of intersectional minority stress among sexual and gender diverse Black, Indigenous, and People of Color. Psychol Sex Orientat Gend Divers 2024; doi: 10.1037/sgd0000748

32.

32. Abreu

, Tyler Lefevor

, Barrita

, et al. Intersectional microaggressions, depressive symptoms, and the role of LGBTQ‐specific parental support in a sample of Latinx sexual and gender minority youth. J Adolesc 2023; 95: 584–595; doi: 10.1002/jad.12139

33.

33. Serpas

, García

, and Ahmed

. Intersectional microaggressions and cardiovascular disease risk among LGBTQ people of color. Ann LGBTQ Public Popul Health 2024; 5: 136–154; doi: 10.1891/LGBTQ20220033

34.

34. Bauer

and Scheim

. Advancing quantitative intersectionality research methods: Intracategorical and intercategorical approaches to shared and differential constructs. Soc Sci Med 2019;226:260–262; doi: 10.1016/j.socscimed.2019.03.018

35.

35. Salerno

, Lea

, and Alcántara

. Effects of racist microaggressions and sexual and gender minority stress on mental health among Latinx lesbian, gay, bisexual, transgender, and queer or questioning+ young adults. Health Educ Behav 2024; 51: 691–699; doi: 10.1177/10901981241254068

36.

36. Craig

, Walsh

, and Quinn

. Intersectional microaggressions, sexual identity concealment, and mental health of young Black sexual minority men and transgender women. Arch Sex Behav 2024; 53: 1245–1254; doi: 10.1007/s10508-023-02777-2

37.

37. Zelaya

, DeBlaere

, and Velez

. Psychometric validation and extension of the LGBT People of Color Microaggression Scale with a sample of sexual minority BIPOC college students. Psychol Sex Orientat Gend Divers 2023; 10: 405–412; doi: 10.1037/sgd0000553

38.

38. Huynh

, Bricker

, Lee

, et al. Development and validation of the LGBTQ+ POC Microaggressions Scale—Brief (LGBTQ+ PCMS-B). Stigma Health 2022; 9: 389–399; doi: 10.1037/sah0000419

39.

39. Caba

, Christophe

, Renley

, et al. Measurement invariance of the LGBT People of Color Microaggressions Scale among LGBTQ+ youth. J Res Adolesc 2024; 34: 759–775; doi: 10.1111/jora.12937

40.

40. Huynh

, Balsam

, Abreu

, et al. Development and validation of the LGBTQ+ People of Color Microaggressions Scale-2 (LGBTQ+ PCMS-2). Cultur Divers Ethnic Minor Psychol 2025; doi: 10.1037/cdp0000655

41.

41. Buhrmester

, Kwang

, and Gosling

. Amazon’s mechanical Turk: A new source of inexpensive, yet high-quality, data? Perspect Psychol Sci 2011; 6: 3–5; doi: 10.1177/1745691610393980

42.

42. Peer

, Vosgerau

, and Acquisti

. Reputation as a sufficient condition for data quality on Amazon Mechanical Turk. Behav Res Methods 2014; 46: 1023–1031; doi: 10.3758/s13428-013-0434-y

43.

43.The Williams Institute. Sexual orientation and gender identity (SOGI) adult measures recommendations FAQs. Los Angeles, CA, https://williamsinstitute.law.ucla.edu/wp-content/uploads/SOGI-Measures-FAQ-Mar-2020.pdf [2020, accessed July 30, 2025].

44.

44.Office of Management and Budget (OMB). Revisions to the standards for the classification of federal data on race and ethnicity. Washington, DC, https://obamawhitehouse.archives.gov/omb/fedreg_1997standards [1997, accessed July 30, 2025].

45.

45. Cohen

and Williamson

. Perceived stress in a probability sample of the United States. In: Spacapan

and Oskamp

. (eds.) The Social Psychology of Health. SAGE, 1988, pp. 31–68.

46.

46. Baik

, Fox

, Mills

, et al. Reliability and validity of the perceived stress scale-10 in Hispanic Americans with English or Spanish language preference. J Health Psychol 2019; 24: 628–639; doi: 10.1177/1359105316684938

47.

47. Yılmaz Koğar

and Koğar

. A systematic review and meta‐analytic confirmatory factor analysis of the perceived stress scale (PSS‐10 and PSS‐14). Stress Health 2024; 40: e3285–e13; doi: 10.1002/smi.3285

48.

48.R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/ [2024, accessed July 30, 2025].

49.

49. Rhemtulla

, Brosseau-Liard

PÉ

and Savalei

. When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. Psychol Methods 2012; 17: 354–373; doi: 10.1037/a0029315

50.

50. Beauducel

and Herzberg

. On the performance of maximum likelihood versus means and variance adjusted weighted least squares estimation in CFA. Struct Equ Modeling 2006; 13: 186–203; doi: 10.1207/s15328007sem1302_2

51.

51. Dolan

. Factor analysis of variables with 2, 3, 5 and 7 response categories: A comparison of categorical variable estimators using simulated data. Br J Math Stat Psychol 1994; 47: 309–326; doi: 10.1111/j.2044-8317.1994.tb01039.x

52.

52. Putnick

and Bornstein

. Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Dev Rev 2016;41:71–90; doi: 10.1016/j.dr.2016.06.004

53.

53. Meredith

. Measurement invariance, factor analysis, and factorial invariance. Psychometrika 1993;58:525–543; doi: 10.1007/BF02294825

54.

54. Chen

. Sensitivity of goodness of fit indexes to lack of measurement invariance. Struct Equ Modeling 2007; 14: 464–504; doi: 10.1080/10705510701301834

55.

55. Hoy-Ellis

. Minority stress and mental health: A review of the literature. J Homosex 2023; 70: 806–830; doi: 10.1080/00918369.2021.2004794

56.

56. Hunsley

and Mash

. Evidence-based assessment. Annu Rev Clin Psychol 2007; 3: 29–51; doi: 10.1146/annurev.clinpsy.3.022806.091419

57.

57. Weston

and Gore

PJ.

Jr ., A brief guide to structural equation modeling. Couns Psychol 2006; 34: 719–751; doi: 10.1177/0011000006286345

58.

58. LaMartine

, García

, and Nakamura

. “Scars on your heart and PTSD in your mind”: Transmisogynoir and the well-being of Black transgender women in Los Angeles. J Prev Health Promot 2024; 5: 235–263; doi: 10.1177/26320770241252766

59.

59. Singh

, Bhambhani

, Skinta

, et al. Measurement of intersectional microaggressions: Conceptual barriers and recommendations. Perspect Psychol Sci 2021; 16: 956–971; doi: 10.1177/1745691621991855

60.

Quintana

and Scull

NC.

Latino ethnic identity. In Villarruel

, Carlo

, and Grau

, et al. (eds.) Handbook of US Latino Psychology: Developmental and Community-Based Perspectives. Thousand Oaks, CA: Sage Publications Inc; 2009, pp. 81–98.

61.

61. Flores

, Telles

, Ventura

. New OMB’s race and ethnicity standards will affect how Americans self-identify. Sociol Sci 2024;11:1147–1169; doi: 10.15195/v11.a42

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.01 MB

0.00 MB

0.01 MB

0.02 MB